Global Foundry Market Outlook 2026
Why Leading-Edge Manufacturing Is Becoming More Strategic
iAtlas | Insight | Market Outlook #09

The global foundry market outlook for 2026 shows an industry becoming more concentrated, more capital-intensive and increasingly strategic.
Artificial intelligence is driving extraordinary demand for leading-edge semiconductor manufacturing.
GPUs, custom AI accelerators, CPUs and networking processors require advanced logic processes. HBM is creating additional demand for logic base dies. Advanced packaging requires silicon interposers and other foundry-produced components.
At the same time, mature process nodes remain essential for power management, analog ICs, connectivity, display drivers and thousands of supporting chips.
The result is a foundry market growing at both ends of the technology spectrum.
At the leading edge, 2nm and 3nm capacity has become a strategic resource.
At mature nodes, AI infrastructure and industrial demand are tightening capacity that was once considered abundant.
TrendForce reported that revenue among the world’s ten largest foundries reached approximately $53.49 billion in Q2 2026, up 11.5% quarter over quarter and setting another record.
TSMC alone accounted for 72.5% of the market, followed by Samsung Foundry at 5.9% and SMIC at 5.4%.
Foundry competition is therefore no longer simply about who can manufacture the smallest transistor.
It increasingly involves:
Technology
Yield
Capacity
Advanced Packaging
Supply Security
Geographic Footprint
The foundry business is becoming one of the most important strategic layers in the global semiconductor industry.
Key Takeaways
- Revenue among the world’s top ten foundries reached approximately $53.49 billion in Q2 2026, another record high.
- TSMC maintained a dominant 72.5% market share in Q2 2026.
- TSMC’s 2nm process already represented 8% of wafer revenue in Q3 2026, while 3nm represented 33%.
- Samsung Foundry is expanding 2nm production and reported stronger advanced-node and HBM base-die business during 2026.
- Intel’s 18A process entered production in 2025, while enhanced 18A-P entered risk production in 2026.
- Foundry competition is increasingly connected to AI accelerators, HBM, chiplets and advanced packaging.
- Mature nodes remain strategically important, with strong utilization in areas such as power, analog and AI peripheral chips.
- SEMI expects foundry and logic wafer-fab equipment sales to reach approximately $78 billion in 2026, up 18.9% year over year.
- Geographic diversification is accelerating as the United States, Japan, Europe and other regions seek more local semiconductor capacity.
- The next foundry winners will likely be determined not only by process-node leadership, but by yield, ecosystem, capacity execution and customer trust.
1. The Foundry Market Is Reaching a New Scale
A semiconductor foundry manufactures chips designed by other companies.
This model enabled the growth of fabless companies such as NVIDIA, AMD, Qualcomm, Broadcom and many others.
Instead of building semiconductor factories costing tens of billions of dollars, chip designers can focus on architecture and product development while relying on specialist manufacturers.
For years, this division of labor steadily expanded the foundry industry.
AI is now accelerating it.
TrendForce reported that combined revenue among the world’s top ten foundries reached approximately $53.49 billion in Q2 2026, up 11.5% from the previous quarter.
Several factors contributed simultaneously:
- AI and HPC processor demand
- Advanced-node shortages
- AI-related power and peripheral IC demand
- Consumer inventory replenishment
- Higher mature-node utilization
This is significant because the current cycle is not being driven by one semiconductor category alone.
AI demand is beginning to spread from the most advanced processors toward a much larger supporting chip ecosystem.
2. TSMC’s Dominance Has Become Extraordinary
The most striking feature of the global foundry market is TSMC’s position.
TrendForce estimates that TSMC generated approximately $40.2 billion in foundry revenue in Q2 2026, representing 72.5% of the top-ten foundry market.
The gap between TSMC and every other foundry is enormous.
Samsung Foundry ranked second with approximately 5.9%.
SMIC followed at 5.4%.
That means TSMC’s advantage is no longer simply incremental.
It has become structural.
Several factors reinforce this position:
Leading-Edge Process Technology
↓
Large Customer Base
↓
High Utilization
↓
Manufacturing Learning
↓
Yield Improvement
↓
More Customer Confidence
↓
More Leading-Edge Orders
This creates a powerful cycle.
Customers want access to proven leading-edge manufacturing.
Large customer volumes improve manufacturing experience.
That experience helps improve yield and process stability.
And better manufacturing performance attracts more customers.
3. 2nm Has Already Become Commercially Important
The transition toward 2nm is no longer only a roadmap discussion.
It is generating real revenue.
TSMC reported that 2nm represented 8% of total wafer revenue in Q3 2026, up from 3% in the previous quarter.
During the same quarter:
3nm: 33%
5nm: 28%
7nm: 10%
Processes at 7nm and below together represented 79% of total wafer revenue.
This illustrates how strongly TSMC’s economics are now tied to leading-edge manufacturing.
AI accelerators and high-performance computing devices benefit significantly from advanced process technology because they require:
- Higher transistor density
- Better performance
- Lower power consumption
- More efficient computing
AI therefore increases the economic value of every improvement in process technology.
4. AI Is Increasing the Value of Leading-Edge Capacity
The AI semiconductor boom creates an unusual constraint.
Demand is not simply for more wafers.
It is for very specific advanced wafers.
AI accelerators increasingly require 5nm, 4nm, 3nm and eventually 2nm technologies.
These processes are among the most difficult manufacturing technologies in the world.
Capacity cannot be added instantly.
A new fab requires:
Land
↓
Construction
↓
Cleanrooms
↓
Equipment Installation
↓
Process Qualification
↓
Yield Ramp
↓
High-Volume Manufacturing
This can take years.
As a result, leading-edge foundry capacity increasingly behaves like scarce infrastructure.
TrendForce has described intense AI-driven competition for 3nm and 2nm wafer capacity alongside advanced packaging resources.
For AI companies, securing compute capacity therefore increasingly means securing manufacturing capacity years in advance.
5. Samsung Is Trying to Strengthen Its Second-Source Position
Samsung Foundry remains TSMC’s most visible competitor at the leading edge.
Its market share is much smaller, but Samsung possesses several capabilities that make it strategically important.
The company operates:
- Advanced logic manufacturing
- Memory manufacturing
- HBM
- Advanced packaging
- System LSI design
This gives Samsung an unusual degree of vertical integration.
Samsung reported significant improvement in its Foundry business during Q3 2026, driven partly by increased HBM base-die sales and stronger advanced-node demand.
The company said it continued winning 2nm HPC customers, including large hyperscale customers, while expanding 4nm and 2nm production.
Samsung also plans to use its Taylor, Texas fab to expand 2nm AI/HPC manufacturing.
This creates a strategically important alternative for customers that may not want complete dependence on one foundry.
6. But Catching TSMC Is About More Than Process Technology
Semiconductor competition is often described in terms of node numbers.
3nm vs 3nm
2nm vs 2nm
But identical node labels do not mean identical manufacturing performance.
Customers also evaluate:
- Yield
- Power characteristics
- Performance
- Design ecosystem
- IP availability
- Production stability
- Capacity
- Delivery reliability
- Packaging availability
Yield is particularly important.
A technically excellent process becomes economically unattractive if too many chips on each wafer fail.
For large AI processors, this becomes even more critical because large dies are expensive and sensitive to manufacturing defects.
The leading-edge foundry race is therefore not simply a race to announce a process first.
It is a race to manufacture that process economically at scale.
7. Intel Is Trying to Re-enter the Foundry Competition
Intel represents another important variable.
Historically, Intel manufactured primarily for its own products.
The company is now attempting to build a broader external foundry business.
Intel’s 18A process entered production in 2025.
The technology introduced two major architectural changes:
RibbonFET
Intel’s gate-all-around transistor architecture.
and
PowerVia
A backside power-delivery technology.
In June 2026, Intel announced that 18A-P, an enhanced version of 18A, had entered risk production.
Intel hopes these technologies can attract external customers requiring advanced manufacturing for AI and high-performance computing.
If successful, Intel could become another meaningful leading-edge manufacturing option.
But building a foundry business requires much more than strong transistor technology.
Intel must demonstrate:
Process Technology
Yield
Design Ecosystem
Customer Service
Reliable Manufacturing
The challenge is as commercial as it is technical.
8. SMIC Is Becoming More Important
The foundry market is not only a competition among TSMC, Samsung and Intel.
China’s SMIC is also growing rapidly.
TrendForce reported that SMIC generated more than $3 billion in Q2 2026, up approximately 20% quarter over quarter.
Its market share increased to 5.4%, narrowing the gap with Samsung’s 5.9%.
SMIC benefits from enormous domestic semiconductor demand and China’s broader strategy to localize semiconductor manufacturing.
Its growth is particularly relevant across:
- Consumer ICs
- Power management
- Networking
- AI peripheral chips
- Mature and selected advanced nodes
Technology restrictions may constrain access to some leading-edge equipment.
But Chinese demand creates substantial incentives to expand domestic capacity.
That means SMIC’s strategic importance could continue increasing even if its technology roadmap differs from TSMC’s.
9. Mature Nodes Are Not Becoming Obsolete
Leading-edge technology receives most of the attention.
But the semiconductor industry still depends heavily on older process nodes.
A modern AI server may use 3nm logic for the primary accelerator while simultaneously requiring mature-node chips for:
- Power management
- Voltage regulation
- Analog functions
- Controllers
- Connectivity
- Sensors
- Interfaces
TrendForce reported that AI-related demand is increasingly affecting both advanced and mature foundry capacity.
For example, AI servers require large numbers of PMICs and power devices.
Those chips often use mature processes.
This creates an important distinction:
Advanced AI systems do not eliminate mature nodes.
They can actually create more demand for them.
10. UMC Shows the Strength of Mature-Node Demand
UMC is an important example.
The company does not compete directly with TSMC at the most advanced nodes.
Instead, it focuses heavily on mature and specialty processes.
In Q3 2026, UMC reported:
- Revenue of approximately $2.32 billion
- 92% capacity utilization
- 22/28nm representing approximately 35% of wafer revenue
The company also reported record wafer shipments during the quarter.
This demonstrates that mature foundry manufacturing can remain highly valuable.
Not every semiconductor requires 2nm.
In many applications, mature processes can provide better economics, proven reliability and sufficient performance.
11. GlobalFoundries Is Following a Differentiated Strategy
GlobalFoundries follows another model.
Rather than competing directly for 2nm leadership, the company focuses on specialized semiconductor technologies.
These include applications such as:
- Automotive
- RF
- Silicon photonics
- Power
- Connectivity
- Industrial
- Aerospace and defense
In Q3 2026, GlobalFoundries reported $1.899 billion in revenue and highlighted particularly strong momentum in communications infrastructure, data centers and automotive applications.
Its AI opportunity is also expanding beyond processors.
GlobalFoundries is investing in technologies used in:
Optical Connectivity
Power Management
Silicon Interposers
These are increasingly important components of AI infrastructure.
This reinforces a broader point:
The foundry market has multiple competitive models.
Not every successful foundry needs to manufacture the most advanced CPU or GPU.
12. Foundry Competition Is Expanding Into Advanced Packaging
The boundary between foundry and packaging is becoming increasingly blurred.
AI chips require close integration between:
Logic
HBM
Chiplets
Interposers
Advanced Packaging
TSMC has built a powerful advantage through its combination of leading-edge manufacturing and CoWoS packaging.
Samsung is similarly trying to combine logic, HBM and packaging.
Intel is integrating its foundry strategy with EMIB and Foveros.
Even GlobalFoundries is expanding silicon-interposer capabilities for AI packaging applications.
The implication is significant.
Future foundry competition may increasingly be based on a combined offering:
Wafer Manufacturing
Advanced Packaging
rather than wafer manufacturing alone.
This is why the topics covered in Market Outlook #06 HBM and #07 Advanced Packaging connect directly to the foundry market.
13. Foundry Investment Is Accelerating
These trends require enormous capital.
SEMI forecasts wafer-fab equipment sales for foundry and logic applications to reach approximately $78 billion in 2026, an increase of 18.9% year over year.
Growth is expected to continue as the industry expands:
- 2nm GAA capacity
- AI accelerator production
- HPC manufacturing
- Advanced-node mobile processors
SEMI’s latest 300mm Fab Outlook also projects global front-end 300mm fab equipment spending of approximately $149 billion in 2026, up 31% year over year.
Logic & Micro — including foundry — is expected to remain the largest investment segment.
The foundry race is therefore also a semiconductor-equipment investment race.
14. The Foundry Industry Is Becoming More Geographic
For decades, semiconductor manufacturing concentrated heavily in East Asia because specialization and manufacturing scale created enormous economic efficiencies.
Those economics remain powerful.
But governments increasingly view semiconductor manufacturing as strategic infrastructure.
This is driving investment across:
Taiwan
South Korea
United States
Japan
Europe
China
TSMC is expanding outside Taiwan.
Samsung is investing in the United States.
Intel is building a broader manufacturing network.
GlobalFoundries is expanding U.S. capacity.
Japan is rebuilding parts of its semiconductor-manufacturing ecosystem.
SEMI projects continued construction of new 300mm fabs worldwide through the remainder of the decade.
This does not mean semiconductor manufacturing will become fully localized.
The ecosystem is too interconnected.
But the geographic structure is becoming more diversified.
15. Regionalization Comes With a Cost
Manufacturing semiconductors in multiple regions can improve supply-chain resilience.
But duplication is expensive.
Leading-edge fabs require:
- Tens of billions of dollars in investment
- Massive cleanroom infrastructure
- Specialized engineers
- Large quantities of electricity
- Ultra-pure water
- Global equipment suppliers
- Complex chemical and material supply chains
A fab cannot operate independently simply because it is built domestically.
Equipment may come from the United States, Europe or Japan.
Materials may come from Japan, Korea or other countries.
Design IP may come from another region.
Advanced packaging may occur somewhere else.
The semiconductor supply chain therefore remains global even as manufacturing becomes more regional.
This creates a new optimization problem:
Efficiency
versus
Resilience
The foundry industry will increasingly need to manage both.
16. Power and Water Are Becoming Strategic Constraints
Semiconductor fabs are massive industrial facilities.
Leading-edge production requires enormous amounts of:
Electricity
and
Ultra-Pure Water
The more fabs the world builds, the more important these physical constraints become.
AI adds another layer.
The same AI boom driving semiconductor manufacturing investment is also driving enormous electricity demand from data centers.
That means fabs and data centers can increasingly compete within the same infrastructure environment for:
- Electricity
- Grid connections
- Land
- Construction resources
- Skilled workers
Semiconductor manufacturing therefore cannot be analyzed separately from energy and infrastructure.
The next stage of foundry expansion depends not only on semiconductor demand but also on whether regions can physically support new fabs.
17. The Economics of Foundry Are Becoming More Difficult
Leading-edge semiconductor manufacturing has become extraordinarily expensive.
Each new process generation can require:
- New fabrication equipment
- More lithography steps
- More complex process control
- New transistor structures
- New materials
- Larger R&D budgets
This increases the importance of utilization.
A semiconductor fab has enormous fixed costs.
If utilization is high, those costs are spread across more wafers.
If utilization falls, profitability can deteriorate quickly.
This creates one of the core risks of the foundry cycle.
Strong Demand
↓
Massive Capacity Investment
↓
More Fabs
↓
Potential Oversupply
↓
Lower Utilization
↓
Margin Pressure
AI currently supports exceptionally strong investment.
But semiconductor history suggests that capacity cycles remain important.
18. Foundry Customers Are Becoming More Powerful
Another structural change is happening on the customer side.
The largest AI customers are no longer only traditional chip designers.
Hyperscalers increasingly design their own silicon.
This includes companies developing:
- AI accelerators
- CPUs
- Networking processors
- Custom ASICs
These customers can bring enormous wafer volumes.
They can also demand:
- Custom process optimization
- Guaranteed capacity
- Advanced packaging
- Long-term supply commitments
This strengthens the relationship between foundry and customer.
Future foundry competition may increasingly resemble strategic co-development rather than simple wafer purchasing.
The foundry becomes part of the customer’s technology roadmap.
19. The Foundry Market Is Splitting Into Different Competitive Layers
It is increasingly difficult to describe foundries as one homogeneous industry.
The market is dividing into several layers.
Leading-Edge Foundry
TSMC
Samsung
Intel
Focused on advanced logic and AI/HPC.
Advanced / Large-Scale Chinese Foundry
SMIC
Serving rapidly expanding domestic demand.
Mature-Node Foundry
UMC
Hua Hong
PSMC and others
Focused on cost-efficient, high-volume mature processes.
Specialty Foundry
GlobalFoundries
Tower and others
Focused on differentiated processes such as RF, photonics, power and automotive.
Each group competes differently.
The most advanced node is not necessarily the most important metric in every market.
20. What Could Change the Outlook?
The foundry outlook remains strong, but several risks could alter the trajectory.
AI CapEx Slowdown
If hyperscaler AI investment slows sharply, demand for advanced wafers could weaken.
Overcapacity
Massive global fab construction could eventually create excess capacity.
Yield Challenges
2nm and future nodes require difficult process transitions.
Poor yield could limit effective capacity and profitability.
Geopolitics
Export controls and technology restrictions can reshape where semiconductor equipment and advanced chips are produced.
Infrastructure Constraints
Electricity, water and construction limitations could delay new fabs.
Customer Concentration
Leading-edge foundries increasingly depend on a relatively small number of massive customers.
Rising Manufacturing Cost
The economics of each new process generation are becoming more demanding.
These factors mean that strong semiconductor demand does not automatically guarantee high returns for every foundry.
iAtlas Outlook
The global foundry market outlook shows a semiconductor industry becoming increasingly strategic.
Foundry Market 1.0
Fabless Growth
↓
Wafer Outsourcing
↓
Process Scaling
↓
Cost & Yield Competition
The next phase is much broader.
Foundry Market 2.0
AI + HPC
↓
2nm & Advanced Nodes
↓
HBM Base Dies
↓
Advanced Packaging
↓
Massive Fab Investment
↓
Regional Supply Chains
↓
Strategic Manufacturing Capacity
The key shift is that foundry capacity is no longer simply semiconductor production capacity.
At the leading edge, it is becoming computing infrastructure.
TSMC’s position demonstrates how powerful this can become.
Its combination of process technology, yield, capacity, customer relationships and advanced packaging has created an unusually strong competitive position.
Samsung is attempting to build an alternative through 2nm, HBM and packaging integration.
Intel is trying to re-establish itself as a major external foundry through 18A.
SMIC continues to expand alongside China’s semiconductor localization strategy.
Meanwhile, UMC, GlobalFoundries and other companies demonstrate that mature and specialty nodes remain highly valuable.
The market is therefore not converging toward one manufacturing model.
It is becoming more segmented.
But across every segment, one factor is becoming more important:
manufacturing capability itself is becoming strategic.
The next foundry winners will not simply be the companies that announce the smallest process node.
They will be the companies that can deliver:
Technology
Yield
Capacity
Packaging
Reliability
Supply Security
at the scale demanded by the AI era.
That is why leading-edge manufacturing — and the foundry industry around it — is becoming more strategic.
Continue Reading
Previous Market Outlook
AI Semiconductor Market Outlook 2026
Beyond GPUs: How AI Is Reshaping the Chip Industry
Next Market Outlook
Semiconductor Equipment Market Outlook 2026
Why the AI Investment Cycle Is Expanding Beyond Chips
References
- TrendForce — Global Foundry Revenue Approaches US$53.49 Billion in 2Q26
Global foundry revenue, market share and AI-related foundry demand. - TSMC — Q3 2026 Results
2nm, 3nm and leading-edge wafer revenue data. - Samsung Electronics — Q3 2026 Results
Samsung Foundry 2nm, AI/HPC and Taylor fab strategy. - Intel Foundry — 18A and 18A-P Process Update
Intel’s leading-edge foundry technology roadmap. - SEMI — Global Semiconductor Equipment Forecast 2026–2028
Foundry and logic wafer-fab equipment investment. - SEMI — 300mm Fab Outlook
Global fab construction and equipment-investment outlook. - UMC — Q3 2026 Financial Results
Mature-node utilization and 22/28nm business. - GlobalFoundries — Q3 2026 Financial Results
Specialty foundry, AI infrastructure and capacity strategy.
ℹ️ About iAtlas
iAtlas is an independent publication covering batteries, semiconductors, OLED, advanced materials, AI, and global industrial trends.
We transform complex industrial developments into clear, reliable, and easy-to-understand insights.
Whether you’re following today’s industry news or building long-term expertise, iAtlas helps you understand not only what happened, but why it matters.
Technology creates change.
Insight creates opportunity.
— iAtlas






